StockVane
Four days ago I told you that I was sell put on Baidu because I felt confident that the stock was balanced, the AI pivot was in place, and the strike price was pretty low in comparison to the 52-week low. Review my analysis of Baidu in August. Yesterday Baidu reported Q2 earnings. The share price plunged as much as 13% during the day. My phone rang me with an alert saying the price has dropped, so much so that I thought my portfolio tracker got hacked by someone who has a beef with Chinese tech stocks. It was not hacked. I’m going to go through each of those numbers, explain exactly why the market reacted the way it did and then be honest with you about what this means for the position that I’m in.
What Actually Happened In The Numbers
On August 18, Baidu released its second quarter 2026 results, which saw the top line and bottom line fall short of expectations. Revenue was RMB 31.3 billion (about $4.62 billion), a 4% decrease from the previous year and slightly below the consensus range of $4.65 billion to $4.74 billion, depending on the analyst panel you are looking at. Earnings of $1.06 per ADS were a big disappointment, falling short of the range of analysts’ expectations of $1.35 to $1.46 per ADS, and depending on the way you look at it, a 26.6% shortfall in EPS in local currency terms.
The real story, and the reason why the stock was hammered, not dinged, is in the segment breakdown. The company’s core search advertising business, Online Marketing Services, which has made this company a household name among Chinese internet users for 20 years, dropped 19% YoY to RMB 13.1 billion. It’s not just a quarter-time bump. Several sources monitoring this line item are now reporting that it is the fourth or fifth straight quarter of loss for that business, depending on which base you use for the comparison. The shift from traditional search to AI chatbots and short video platforms is reflected in the numbers and it’s becoming difficult to spin this shift as a mere temporary phenomenon.
General Business revenue, Baidu’s overall operating segment, decreased 4% to RMB 25.2 billion. The streaming arm, iQIYI, fell 5% to RMB 6.3 billion, but the drop was not significant enough to prevent a slight quarter-over-quarter increase.
The AI Story is really powerful, and that’s what makes this complicated
I want to take a breath here and the convenient story the market seems to be telling right now is “Baidu is dying, sell everything,” but I don’t think that’s a story that’s being told by the data.
The revenue of AI-powered businesses rose by 25% YoY to RMB 12.5 billion. The revenue of AI Cloud Infrastructure rose by 50% to RMB 7.3 billion. The part of revenue that leases compute power for AI training and inference, GPU Cloud, increased by a whopping 283%. Well, that’s not an insignificant number, it’s a number that could actually be an explosive number, right next to a number that could actually be an alarming number, in the same earnings report, and the market’s job right now is to see which one wins the race first.
The advertising pressure, as described by the management’s own terminology, is expected to continue in the second half of the year, according to the management, due to the high competition for user attention and the industry’s transition to new forms of AI products. That was the closest to an admission you will hear on an earnings call that this is not a blip, but a structural shift that the company is actively going through.
There was also a particular operational fact that I noticed in the call. Apollo Go, Baidu’s robotaxi business, had its ride volume noted as “affected by operational adjustments in some domestic cities due to regulatory considerations,” for the time being. This is corporate speak for “local governments forced us to adapt the way we do business in some markets,” and it’s a reminder that while autonomous ride-hailing is the most futuristic growth vector for Baidu, it’s not just about execution risk that every emerging business line comes with, it’s also about regulatory risk.
Why The Stock Fell As Hard As It Did, Not Just That It Fell
A 4% revenue miss and a significant EPS shortfall are not sufficient indicators of a single-day 13% drawdown, in isolation. The pattern that lies beneath the headline numbers is what justifies that type of reaction. Fitch actually downgraded Baidu’s credit rating from A to A- and it doesn’t happen lightly by credit rating agencies, that’s their way of saying that it’s a permanent change in the earnings base, not a cyclical dip.
It’s an interesting reaction split on the sell side that I think should be walked through rather than being taken in as one headline. Barclays has lowered its price estimate to $124. The Bank of America trimmed its target to $165, but it made it very clear it still had a positive stance on the stock: Buy. That’s a significant message, as analysts continue to reduce their numbers, that there’s a significant group that still feels that the price is still too high for the actual deterioration in the business. One valuation estimate I found valued fair value at $132, nearly 27% up from the crash price and the wider average analyst price is closer to $166, which is over 60% upside from the crash price.
The current position of the stock
The stock was trading in a fairly wide band through the session, opening down around 7% in pre-market, at one point dropping 13% intraday, and closing around $104, but dropping to $90 to $96 at the time of your check, all thanks to the trading action. When zooming out to the full year, however, the shares of Baidu are down by about 35% in 2026, a truly harsh year for a stock that was beginning the year with a lot of optimism behind its AI story.
Forward valuations, even after the earnings disappointment, are still low by historical standards. Forward P/E is in the ballpark of 14.1x and EV/EBITDA is around 3.4x. Those are figures that would seem almost competitively priced in a healthy, growing business. This is either a risk assessment that the market is doing or an overcorrection, and reasonable people with the same numbers are on different sides of that question right now.
Evaluation of this position, step by step
I’m not going to tell you my conclusion, because I think the process is more important than my conclusion, so I want to give you the actual framework that I used last night sitting with this report.
First, break out the headline miss from the segment-level story. A revenue and EPS miss is almost meaningless on its own when it comes to determining if the underlying thesis has changed. You need to go into the real segment breakdown and determine which specific lines of business have gone down and whether that is reflected in your thesis or if it’s new information.
Second, look to see if there has been a change in the management’s own forward-looking statements. In an explicit statement, Baidu’s leadership is predicting that advertising pressure will continue in the second half of this year. This is new information, not a boilerplate caution. If management makes an explicit negative forward statement, rather than making a general statement of optimism, listen.
Third, consider what is accelerating and what is decelerating, and compare the growth rates directly, not the dollar amounts. The growth rate of GPU Cloud is 283% on a relatively small base, while the advertising base is down 19% on a much larger base – that’s a real horse race, not just some feeling of being generally optimistic or pessimistic about “the AI story.
Fourth, review your initial entry logic rather than your overall thoughts on the stock in the present. I sold my put with the strike price of $80 because it was below the 52-week low and a breakeven at $74. The stock, despite this harsh reaction to earnings, remains significantly above both of those levels, in the $90 to $96 area. My original thesis was not “Baidu’s advertising business will continue to thrive forever,” but rather “Baidu’s balance sheet and AI optionality make it worth holding at 25-30% off mid-August prices. Unfortunately, this one is being put to the test, not disproved, of the options thesis-testing model. You can try my Options Profit Calculator before you sell put it.
What I really think about the put position at this moment
I’m not going to lie, this was not a fun 24 hours. When your stock that you have real strike price exposure drops 13% in a day, it causes a very particular stomach ache that can’t be protected against by saying, “I calculated my breakeven ahead of time. In fact, I checked my position more times on yesterday afternoon than I would like to remember — and that’s what I’ve told you, in previous columns, to avoid doing with overly concentrated positions.
However, that is where the discipline pays off. My strike is $80. It was not even close to that level, even at its worst yesterday, which was the low $90s. My delta on this position was last time I checked it before the earnings drop in that 0.15 to 0.20 range, which I’ve mentioned many times, and it’s risen a little bit with the price drop, but still a pretty nice cushion between what I have to pay and what the price is.
What this earnings report has done is to focus my specific risk, which I told you I was taking when I put it down. I made it clear that there is real China-specific risk and structural risk that Baidu faces, and it is not a straightforward US equity analysis. As I said on this call, the regulatory issues with Apollo and the structural issues with advertising are the very things that risk was betting on. I’m not surprised that there is bad news, I am processing what the bad news is.
The following is an excerpt from my book, What I’m Actually Doing From Here
I’m not touching the position. The stock is well clear of both my breakeven and strike price, and my breakeven was established at $74, my strike at $80, and the stock has suffered a double-digit percentage drawdown due to real, disappointing segment data. Either to panic-buy back that put at a loss, or irrationally double down to “average into conviction,” is the type of emotional decision making that I’ve written about several times now and don’t want to be a cautionary tale in my own column.
I am looking at two things in particular over the next two quarters and not the stock price day to day. But first, whether GPU Cloud’s growth rate of 283% will continue or slow down dramatically, because that’s the one number that will make or break the viability of the AI segment being able to replace the advertising slowdown in a timeframe that’s relevant to my June 2027 expiration. Second, if management’s “persist through the second half” rhetoric on advertising pressure becomes a reality – a floor that forms in that segment alters the entire risk equation on this position – or if it fails to stabilize or even worsens into next year, the impact on this position becomes much more complex.
I still think that it’s a pretty solid stock to own at an effective $74 cost basis, as the balance sheet is strong and the growth of AI Cloud is really quite explosive next to the decline in ads. That belief was put to the test yesterday. It didn’t get struck down. It had a bit more of an intricate, slightly more uncomfortable, and definitely more informative upgrade, which is really the best one could hope for after having worked out a strike price they can actually stomach.
